support specify model_type in config

such as norm_train and to_static_train

test_tipc/benchmark_train.sh

@@ -90,6 +90,8 @@ line_num=`expr $line_num + 1`
 fp_items=$(func_parser_value "${lines[line_num]}")
 line_num=`expr $line_num + 1`
 epoch=$(func_parser_value "${lines[line_num]}")
+line_num=`expr $line_num + 1`
+model_type=$(func_parser_value "${lines[line_num]}")
 
 line_num=`expr $line_num + 1`
 profile_option_key=$(func_parser_key "${lines[line_num]}")
@@ -118,6 +120,7 @@ line_gpuid=4
 line_precision=6
 line_epoch=7
 line_batchsize=9
+line_model_type=15
 line_profile=13
 line_eval_py=24
 line_export_py=30
@@ -138,6 +141,7 @@ if [[ ! -n "$PARAMS" ]];then
     batch_size_list=(${batch_size})
     fp_items_list=(${fp_items})
     device_num_list=(N1C4)
+    model_type_list=(${model_type})
     run_mode="DP"
 elif [[ ${PARAMS} = "dynamicTostatic" ]];then
     IFS="|"
@@ -165,14 +169,13 @@ else
     fp_items_list=($precision)
     batch_size_list=($batch_size)
     device_num_list=($device_num)
-fi
 
-# for log name
+    # parse "to_static" options and modify trainer into "to_static_trainer"
-to_static=""
+    if [[ ${model_type} = "dynamicTostatic" ]];then
-# parse "to_static" options and modify trainer into "to_static_trainer"
+        model_type_list="to_static_train"
-if [[ ${model_type} = "dynamicTostatic" ]];then
+    else
-    to_static="d2sT_"
+        model_type_list="norm_train"
-    sed -i 's/trainer:norm_train/trainer:to_static_train/g' $FILENAME
+    fi
 fi
 
 
@@ -180,131 +183,142 @@ IFS="|"
 for batch_size in ${batch_size_list[*]}; do
     for precision in ${fp_items_list[*]}; do
         for device_num in ${device_num_list[*]}; do
-            # sed batchsize and precision
+            for model_type in ${model_type_list[*]}; do
-            func_sed_params "$FILENAME" "${line_precision}" "$precision"
+                # sed batchsize and precision
-            func_sed_params "$FILENAME" "${line_batchsize}" "$batch_size"
+                func_sed_params "$FILENAME" "${line_precision}" "$precision"
-            func_sed_params "$FILENAME" "${line_epoch}" "$epoch"
+                func_sed_params "$FILENAME" "${line_batchsize}" "$batch_size"
-            gpu_id=$(set_gpu_id $device_num)
+                func_sed_params "$FILENAME" "${line_epoch}" "$epoch"
+                func_sed_params "$FILENAME" "${line_model_type}" "$model_type"
 
-            # It is needed that using dali, NHWC and 4 channels when training ResNet50 with AMPO2
+                # for log name
-            if [[ $model_name == "ResNet50" && $precision == "fp16" ]]; then
+                if [[ ${model_type} = "to_static_train" ]];then
-                sed -i "s/ResNet50.yaml/ResNet50_amp_O2_ultra.yaml/g" $FILENAME
+                    to_static="d2sT_"
-            fi
-
-            # if bs is big, then copy train_list.txt to generate more train log
-            # At least 25 log number would be good to calculate ips for benchmark system.
-            # So the copy number for train_list is as follows:
-            total_batch_size=`echo $[$batch_size*${device_num:1:1}*${device_num:3:3}]`
-            if [[ $model_name == *GeneralRecognition* ]]; then
-                cd dataset/
-                train_list_length=`cat train_reg_all_data.txt | wc -l`
-                copy_num=`echo $[25*10*$total_batch_size/$train_list_length]`
-                if [[ $copy_num -gt 1 ]];then
-                    rm -rf train_reg_all_data.txt
-                    for ((i=1; i <=$copy_num; i++));do
-                        cat tipc_shitu_demo_data/demo_train.txt >> train_reg_all_data.txt
-                    done
-                fi
-                cd ..
-            else
-                cd dataset/ILSVRC2012
-                val_list_length=`cat val_list.txt | wc -l`
-                copy_num=`echo $[25*10*$total_batch_size/$val_list_length]`
-                rm -rf train_list.txt
-                if [[ $copy_num -gt 1 ]];then
-                    for ((i=1; i <=$copy_num; i++));do
-                        cat val_list.txt >> train_list.txt
-                    done
                 else
-                    ln -s val_list.txt train_list.txt
+                    to_static=""
                 fi
-                cd ../../
-            fi
 
-            if [[ ${#gpu_id} -le 1 ]];then
+                gpu_id=$(set_gpu_id $device_num)
-                log_path="$SAVE_LOG/profiling_log"
-                mkdir -p $log_path
-                log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}profiling"
-                func_sed_params "$FILENAME" "${line_gpuid}" "0"  # sed used gpu_id
-                # set profile_option params
-                tmp=`sed -i "${line_profile}s/.*/${profile_option}/" "${FILENAME}"`
 
-                # run test_train_inference_python.sh
+                # It is needed that using dali, NHWC and 4 channels when training ResNet50 with AMPO2
-                cmd="bash test_tipc/test_train_inference_python.sh ${FILENAME} benchmark_train > ${log_path}/${log_name} 2>&1 "
+                if [[ $model_name == "ResNet50" && $precision == "fp16" ]]; then
-                echo $cmd
+                    sed -i "s/ResNet50.yaml/ResNet50_amp_O2_ultra.yaml/g" $FILENAME
-                eval $cmd
+                fi
-                eval "cat ${log_path}/${log_name}"
 
-                # without profile
+                # if bs is big, then copy train_list.txt to generate more train log
-                log_path="$SAVE_LOG/train_log"
+                # At least 25 log number would be good to calculate ips for benchmark system.
-                speed_log_path="$SAVE_LOG/index"
+                # So the copy number for train_list is as follows:
-                mkdir -p $log_path
+                total_batch_size=`echo $[$batch_size*${device_num:1:1}*${device_num:3:3}]`
-                mkdir -p $speed_log_path
+                if [[ $model_name == *GeneralRecognition* ]]; then
-                log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}log"
+                    cd dataset/
-                speed_log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}speed"
+                    train_list_length=`cat train_reg_all_data.txt | wc -l`
-                func_sed_params "$FILENAME" "${line_profile}" "null"  # sed profile_id as null
+                    copy_num=`echo $[25*10*$total_batch_size/$train_list_length]`
-                cmd="bash test_tipc/test_train_inference_python.sh ${FILENAME} benchmark_train > ${log_path}/${log_name} 2>&1 "
+                    if [[ $copy_num -gt 1 ]];then
-                echo $cmd
+                        rm -rf train_reg_all_data.txt
-                job_bt=`date '+%Y%m%d%H%M%S'`
+                        for ((i=1; i <=$copy_num; i++));do
-                eval $cmd
+                            cat tipc_shitu_demo_data/demo_train.txt >> train_reg_all_data.txt
-                job_et=`date '+%Y%m%d%H%M%S'`
+                        done
-                export model_run_time=$((${job_et}-${job_bt}))
+                    fi
-                eval "cat ${log_path}/${log_name}"
+                    cd ..
+                else
+                    cd dataset/ILSVRC2012
+                    val_list_length=`cat val_list.txt | wc -l`
+                    copy_num=`echo $[25*10*$total_batch_size/$val_list_length]`
+                    rm -rf train_list.txt
+                    if [[ $copy_num -gt 1 ]];then
+                        for ((i=1; i <=$copy_num; i++));do
+                            cat val_list.txt >> train_list.txt
+                        done
+                    else
+                        ln -s val_list.txt train_list.txt
+                    fi
+                    cd ../../
+                fi
 
-                # parser log
+                if [[ ${#gpu_id} -le 1 ]];then
-                _model_name="${model_name}_bs${batch_size}_${precision}_${run_mode}"
+                    log_path="$SAVE_LOG/profiling_log"
-                cmd="${python} ${BENCHMARK_ROOT}/scripts/analysis.py --filename ${log_path}/${log_name} \
+                    mkdir -p $log_path
-                        --speed_log_file '${speed_log_path}/${speed_log_name}' \
+                    log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}profiling"
-                        --model_name ${_model_name} \
+                    func_sed_params "$FILENAME" "${line_gpuid}" "0"  # sed used gpu_id
-                        --base_batch_size ${batch_size} \
+                    # set profile_option params
-                        --run_mode ${run_mode} \
+                    tmp=`sed -i "${line_profile}s/.*/${profile_option}/" "${FILENAME}"`
-                        --fp_item ${precision} \
-                        --keyword ips: \
-                        --skip_steps 100 \
-                        --device_num ${device_num} \
-                        --speed_unit samples/s \
-                        --convergence_key loss: "
-                echo $cmd
-                eval $cmd
-                last_status=${PIPESTATUS[0]}
-                status_check $last_status "${cmd}" "${status_log}" "${model_name}"
-            else
-                IFS=";"
-                unset_env=`unset CUDA_VISIBLE_DEVICES`
-                log_path="$SAVE_LOG/train_log"
-                speed_log_path="$SAVE_LOG/index"
-                mkdir -p $log_path
-                mkdir -p $speed_log_path
-                log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}log"
-                speed_log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}speed"
-                func_sed_params "$FILENAME" "${line_gpuid}" "$gpu_id"  # sed used gpu_id
-                func_sed_params "$FILENAME" "${line_profile}" "null"  # sed --profile_option as null
-                cmd="bash test_tipc/test_train_inference_python.sh ${FILENAME} benchmark_train > ${log_path}/${log_name} 2>&1 "
-                echo $cmd
-                job_bt=`date '+%Y%m%d%H%M%S'`
-                eval $cmd
-                job_et=`date '+%Y%m%d%H%M%S'`
-                export model_run_time=$((${job_et}-${job_bt}))
-                eval "cat ${log_path}/${log_name}"
-                # parser log
-                _model_name="${model_name}_bs${batch_size}_${precision}_${run_mode}"
 
-                cmd="${python} ${BENCHMARK_ROOT}/scripts/analysis.py --filename ${log_path}/${log_name} \
+                    # run test_train_inference_python.sh
-                        --speed_log_file '${speed_log_path}/${speed_log_name}' \
+                    cmd="bash test_tipc/test_train_inference_python.sh ${FILENAME} benchmark_train > ${log_path}/${log_name} 2>&1 "
-                        --model_name ${_model_name} \
+                    echo $cmd
-                        --base_batch_size ${batch_size} \
+                    eval $cmd
-                        --run_mode ${run_mode} \
+                    eval "cat ${log_path}/${log_name}"
-                        --fp_item ${precision} \
+
-                        --keyword ips: \
+                    # without profile
-                        --skip_steps 100 \
+                    log_path="$SAVE_LOG/train_log"
-                        --device_num ${device_num} \
+                    speed_log_path="$SAVE_LOG/index"
-                        --speed_unit images/s \
+                    mkdir -p $log_path
-                        --convergence_key loss: "
+                    mkdir -p $speed_log_path
-                echo $cmd
+                    log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}log"
-                eval $cmd
+                    speed_log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}speed"
-                last_status=${PIPESTATUS[0]}
+                    func_sed_params "$FILENAME" "${line_profile}" "null"  # sed profile_id as null
-                status_check $last_status "${cmd}" "${status_log}" "${model_name}"
+                    cmd="bash test_tipc/test_train_inference_python.sh ${FILENAME} benchmark_train > ${log_path}/${log_name} 2>&1 "
-            fi
+                    echo $cmd
+                    job_bt=`date '+%Y%m%d%H%M%S'`
+                    eval $cmd
+                    job_et=`date '+%Y%m%d%H%M%S'`
+                    export model_run_time=$((${job_et}-${job_bt}))
+                    eval "cat ${log_path}/${log_name}"
+
+                    # parser log
+                    _model_name="${model_name}_bs${batch_size}_${precision}_${run_mode}"
+                    cmd="${python} ${BENCHMARK_ROOT}/scripts/analysis.py --filename ${log_path}/${log_name} \
+                            --speed_log_file '${speed_log_path}/${speed_log_name}' \
+                            --model_name ${_model_name} \
+                            --base_batch_size ${batch_size} \
+                            --run_mode ${run_mode} \
+                            --fp_item ${precision} \
+                            --keyword ips: \
+                            --skip_steps 100 \
+                            --device_num ${device_num} \
+                            --speed_unit samples/s \
+                            --convergence_key loss: "
+                    echo $cmd
+                    eval $cmd
+                    last_status=${PIPESTATUS[0]}
+                    status_check $last_status "${cmd}" "${status_log}" "${model_name}"
+                else
+                    IFS=";"
+                    unset_env=`unset CUDA_VISIBLE_DEVICES`
+                    log_path="$SAVE_LOG/train_log"
+                    speed_log_path="$SAVE_LOG/index"
+                    mkdir -p $log_path
+                    mkdir -p $speed_log_path
+                    log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}log"
+                    speed_log_name="${repo_name}_${model_name}_bs${batch_size}_${precision}_${run_mode}_${device_num}_${to_static}speed"
+                    func_sed_params "$FILENAME" "${line_gpuid}" "$gpu_id"  # sed used gpu_id
+                    func_sed_params "$FILENAME" "${line_profile}" "null"  # sed --profile_option as null
+                    cmd="bash test_tipc/test_train_inference_python.sh ${FILENAME} benchmark_train > ${log_path}/${log_name} 2>&1 "
+                    echo $cmd
+                    job_bt=`date '+%Y%m%d%H%M%S'`
+                    eval $cmd
+                    job_et=`date '+%Y%m%d%H%M%S'`
+                    export model_run_time=$((${job_et}-${job_bt}))
+                    eval "cat ${log_path}/${log_name}"
+                    # parser log
+                    _model_name="${model_name}_bs${batch_size}_${precision}_${run_mode}"
+
+                    cmd="${python} ${BENCHMARK_ROOT}/scripts/analysis.py --filename ${log_path}/${log_name} \
+                            --speed_log_file '${speed_log_path}/${speed_log_name}' \
+                            --model_name ${_model_name} \
+                            --base_batch_size ${batch_size} \
+                            --run_mode ${run_mode} \
+                            --fp_item ${precision} \
+                            --keyword ips: \
+                            --skip_steps 100 \
+                            --device_num ${device_num} \
+                            --speed_unit images/s \
+                            --convergence_key loss: "
+                    echo $cmd
+                    eval $cmd
+                    last_status=${PIPESTATUS[0]}
+                    status_check $last_status "${cmd}" "${status_log}" "${model_name}"
+                fi
+            done
         done
     done
 done

test_tipc/configs/CSWinTransformer/CSWinTransformer_tiny_224_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/DenseNet/DenseNet121_train_infer_python.txt

@@ -50,5 +50,12 @@ inference:python/predict_cls.py -c configs/inference_cls.yaml
 -o Global.benchmark:False
 null:null
 null:null
+===========================train_benchmark_params==========================
+batch_size:64
+fp_items:fp32|fp16
+epoch:1
+model_type:norm_train|to_static_train
+--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
+flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================
 random_infer_input:[{float32,[3,224,224]}]

test_tipc/configs/GeneralRecognition/GeneralRecognition_PPLCNet_x2_5_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:256
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/GeneralRecognitionV2/GeneralRecognitionV2_PPLCNetV2_base_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/HRNet/HRNet_W18_C_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:64
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/HRNet/HRNet_W48_C_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:64|128
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/MobileNetV1/MobileNetV1_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:64|128
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/MobileNetV2/MobileNetV2_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:64|128
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/MobileNetV3/MobileNetV3_large_x1_0_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:256|640
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train|to_static_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/MobileNetV3/MobileNetV3_large_x1_0_train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:256|640
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/MobileNetV3/MobileNetV3_large_x1_0_train_pact_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:256|640
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/MobileNetV3/MobileNetV3_large_x1_0_train_ptq_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:256|640
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/MobileViT/MobileViT_S_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPHGNet/PPHGNet_small_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:128
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPHGNet/PPHGNet_tiny_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:128
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPLCNet/PPLCNet_x0_25_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:512
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPLCNet/PPLCNet_x0_35_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:512
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPLCNet/PPLCNet_x0_5_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:512
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPLCNet/PPLCNet_x0_75_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:512
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPLCNet/PPLCNet_x1_0_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:512
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPLCNet/PPLCNet_x1_5_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:512
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPLCNet/PPLCNet_x2_0_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:512
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPLCNet/PPLCNet_x2_5_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:256
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PPLCNetV2/PPLCNetV2_base_train_infer_python.txt

@@ -53,6 +53,7 @@ null:null
 batch_size:500
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/PVTV2/PVT_V2_B2_Linear_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/ResNeXt/ResNeXt50_32x4d_train_infer_python.txt

@@ -50,5 +50,12 @@ inference:python/predict_cls.py -c configs/inference_cls.yaml
 -o Global.benchmark:False
 null:null
 null:null
+===========================train_benchmark_params==========================
+batch_size:64
+fp_items:fp32|fp16
+epoch:1
+model_type:norm_train|to_static_train
+--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
+flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================
 random_infer_input:[{float32,[3,224,224]}]

test_tipc/configs/ResNet/ResNet152_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:32|64
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/ResNet/ResNet50_train_ampfp16_infer_python.txt

@@ -54,3 +54,4 @@ null:null
 batch_size:128|256
 fp_items:ampfp16
 epoch:1
+model_type:norm_train

test_tipc/configs/ResNet/ResNet50_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128|64
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train|to_static_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/ResNet/ResNet50_train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/ResNet/ResNet50_train_linux_gpu_normal_amp_infer_python_linux_gpu_cpu.txt

@@ -54,3 +54,4 @@ null:null
 batch_size:128|256
 fp_items:purefp16
 epoch:1
+model_type:norm_train

test_tipc/configs/ResNet/ResNet50_train_pact_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/ResNet/ResNet50_train_ptq_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/ResNet/ResNet50_vd_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:64
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/ResNet/ResNet50_vd_train_pact_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/ResNet/ResNet50_vd_train_ptq_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/ShuffleNet/ShuffleNetV2_x1_0_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:256|1536
 fp_items:fp32|fp16
 epoch:2
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/SwinTransformer/SwinTransformer_tiny_patch4_window7_224_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:104|128
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train|to_static_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/SwinTransformer/SwinTransformer_tiny_patch4_window7_224_train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:64|104
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/SwinTransformer/SwinTransformer_tiny_patch4_window7_224_train_pact_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:64|104
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/SwinTransformer/SwinTransformer_tiny_patch4_window7_224_train_ptq_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:64|104
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/Twins/alt_gvt_base_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:64|144
 fp_items:fp32
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================

test_tipc/configs/Twins/alt_gvt_small_train_infer_python.txt

@@ -54,6 +54,7 @@ null:null
 batch_size:128
 fp_items:fp32|fp16
 epoch:1
+model_type:norm_train
 --profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=4096
 ===========================infer_benchmark_params==========================